Performance Adviser - How to find Average R/W I/O size

[markryanwillis]

I cannot seem to find this counter anywhere within DFM and have activated all special counters and have created custom views in Performance Adviser. I understand NetApp uses the 4k block for processing at the disk after it leaves the front-end ports, but need to determine what my I/O block-size mix is that hits my array. NetApp indicated the counter should be located under the "Target" counter, but seen nothing that looks like Average R/W Block I/O size. Maybe it's called something else? Anyone?

[joker_morgan]

Hi Mark,

Not sure if this is what you are after, but here goes.  If you run NFS stat -d you get what the calls are and how many in each block size.   By the way... you said that you have activated all special counters... can you send me any info on doing that? I am trying to measure average read and write latency for LUNs and when I try to build the view in the nmc I can see the counters that I want but it says that they are currently disabled.

Anyway... I have been looking into similar things lately and will post if I come across average block size.

Server nfs V3: (10472591952 calls)

null       getattr    setattr    lookup     access     readlink   read

719573 0%  3091300682 30% 9285901 0% 108448744 1% 196642215 2% 492878 0%  3294648946 31%

write      create     mkdir      symlink    mknod      remove     rmdir

3652494337 35% 8117853 0% 78572763 1% 10267 0%   0 0%       5110354 0% 169440 0%

rename     link       readdir    readdir+   fsstat     fsinfo     pathconf

332190 0%  647367 0%  2678763 0% 10186353 0% 11942399 0% 780258 0%  10669 0%

commit

0 0%

Read request stats (version 3)

0-511      512-1023   1K-2047    2K-4095    4K-8191    8K-16383   16K-32767  32K-65535  64K-131071 > 131071

882835     1160717    1239622    1242489    27672958   245727834  10057425   2863000088 143761938  0

Write request stats (version 3)

0-511      512-1023   1K-2047    2K-4095    4K-8191    8K-16383   16K-32767  32K-65535  64K-131071 > 131071

2847114    1878793    2995831    4480958    66931547   38517381   5355888    3008322084 521167461  0

[markryanwillis]

Enable showing of diagnostic counters in DFM: dfm option set perfAdvisorShowDiagCounters=yes

Restart DFM and then create your custom views, charts, etc.

I think there should be a specific counter within DFM that would capture this as it's a pretty standard metric as far as I can tell. Using NFS stat -d does not provide me the average read/write I/O block size coming in from attached FC-host, so not sure that will work for all around analysis of SAN throughput as well as other NAS protocol's such as CIFS I/O request.

[AGUMADAVALLI]

From the Operations manager or latest oncommand core, download the NMC (NetApp Management Console) which will allow you to set the real time counters and can show the past analysis too.

thank you,

AK G

[markryanwillis]

AK G

Not sure if you understood what I had stated, but I'm using NMC 3.1